Selective, bifunctional Cu–WO_x/Al₂O₃ catalyst for hydrodeoxygenation of fatty acids

Abstract

Selective hydrodeoxygenation of oleic acid (OA; in a batch reactor, at 300 °C, 30 bar H₂ pressure, reaction time of 1 h and reactant-to-catalyst weight ratio of 2 g OA/0.2 g) forming n-octadecane in yields as high as 93% over a non-noble metal, sulfur-free, bimetallic Cu–WO_x/Al₂O₃ catalyst is reported for the first time. Several Cu–WO_x/Al₂O₃ compositions were prepared by a sequential wet-impregnation method and evaluated. A catalyst with 10 wt% Cu and 4 wt% W enabled the highest activity and selectivity. Upon adding WO_x, the amount of moderate and strong acid sites increased and more Cu in the catalyst was in reduced electron-rich metallic (Cu⁰) state. The crystallite size and dispersion of Cu were little affected. WO_x promoted the fatty acid hydrodeoxygenation activity of Cu. While a monometallic Cu catalyst (10Cu/Al₂O₃) yielded n-octadecane and n-heptadecane along with high amounts of intermediate octadecanol and octadecanal products, the bimetallic catalyst (10Cu–4WO_x/Al₂O₃) gave mainly n-octadecane. Acidity and the high amount of reduced Cu⁰ species are responsible for the high catalytic hydrodeoxygenation performance of this bimetallic catalyst.

---

Introduction

Fuel-grade hydrocarbons (HCs) produced from inedible vegetable oils, animal fats and fatty acids are a sustainable alternative to fossil-derived transport fuels.^1–4 Their deployment leads to energy independence, reduced greenhouse gas (GHG) emissions, savings on oil import bill and rural empowerment. Their exploitation in transportation aligns with the mandate of several nations towards the use of renewable fuels.⁵ Deoxygenation is the pathway by which these HCs are produced.^6,7 It is achieved by two means: (1) hydrodeoxygenation (HDO), wherein oxygen in fatty acid is taken out as water under a hydrogen environment, and (2) decarboxylation/decarbonylation (DCO), wherein oxygen is removed as CO₂/CO. In the former route (HDO), a HC with carbon chain length the same as that of the reactant fatty compound is produced (for example, oleic acid (OA) is converted into octadecane, C₁₈), while in the latter route (DCO), a HC with carbon chain length shorter by one unit than the original fatty compound is produced (i.e., OA is converted into heptadecane, C₁₇). The renewable HCs thus formed outperform the fossil-derived fuels with cetane number (CN) ranging from 85 to 99 compared to 45 to 55 for the conventional diesel. They can be used as stand-alone fuel in conventional diesel engines after mild hydroisomerization.

Neste Oil, UOP/Eni (Ecofining™), Honeywell (Green Jet Fuel Technology) and ConocoPhillips–Tyson Foods announced commercial processes of renewable HCs using conventional hydrotreating catalysts such as sulfided CoMo and sulfided NiMo supported on Al₂O₃.^8–12 These processes operate at high temperature (325–380 °C) and hydrogen pressure (50–80 bar). Continuous sulfidation of the catalyst by passing a sulfur-containing compound in the reactor is essential for its stable performance.¹³ While the starting fatty compound does not contain sulfur in its composition, the product HC contains it due to sulfur leaching. The DCO : HDO product ratio in this process is often in the range of 69 : 21 to 20 : 80.^14,15

Supported noble metals (Pd and Pt) are the other class of deoxygenation catalysts. They operate at moderate temperature (250–300 °C) and in the absence of hydrogen.^16–18 However, a hydrogen supply (5–10 bar) extends the life of the catalyst, avoiding coke formation on the catalyst surface.¹⁹ The absence of hydrogen leads to side reactions such as cracking and heavier compound formation, especially when unsaturated feeds are employed. Unlike that for the conventional hydrotreating catalysts, DCO is the main reaction pathway (>95%) over the noble metal catalysts.^16–18 Acidic metal oxide-promoted noble metals are more efficient catalysts operating at lower temperatures (180–260 °C).^20–24 Adsorption and activation of the fatty compound on the acidic metal oxide and hydrogenation/hydrogenolysis of the C=O bond on the noble metal selectively lead to the HDO product over these bifunctional catalysts. Limited stability, high cost and restricted availability are some drawbacks of the supported noble metal catalysts that impede their industrial application. Thus, development of an affordable, highly active and stable deoxygenation catalyst is still a challenge.

In view of the above challenges, there were several efforts to develop non-sulfided non-noble metal catalysts including mono-elemental and bi-elemental Cu, Ni and Co.^{14,15,25–29} In general, the reduced (non-sulfided) NiMo catalysts showed inferior activity and HC yields compared to the conventional sulfided NiMo catalysts. By optimizing the synergistic atomic ratio of to 0.8 in the reduced NiMo catalysts (instead of 0.3 in the sulfided NiMo catalysts), vegetable oil conversion close to that of the sulfided NiMo catalyst was achieved.^30,31 But then, the sulfided catalyst (1Ni3Mo–S) gave higher HC yield (sunflower oil conversion = 94%, HC yield = 21 wt% with the distribution of C₁₈ = 13 wt%, C₁₇ = 7.0%, C₁₆ = 0.8%, C₁₅ = 0.4 wt%) than the reduced catalyst (3.5Ni0.5Mo) (sunflower oil conversion = 81%, HC yield = 14.2 wt% with the distribution of C₁₈ = 5.5 wt%, C₁₇ = 8.0%, C₁₆ = 0.3%, C₁₅ = 0.4 wt%). DCO (as compared to HDO) is the predominant deoxygenation pathway over the reduced, non-sulfided catalysts.^30,31 Wu et al.³² compared the performance of ZrO₂-supported Cu and Ni catalysts in decarboxylation of stearic acid at 330 °C after 5 h of run under an inert atmosphere. The yield of C₁₇ was only 8% over 20 wt% Cu/ZrO₂, while it was 30% over 20 wt% Ni/ZrO₂. Berenblyum et al.³³ compared the performance of 5 wt% Ni and Cu catalysts supported on γ-Al₂O₃ and WO₃/ZrO₂ in the deoxygenation of stearic acid under 14 bar hydrogen at 350 °C after 6 h of reaction. 1-Heptadecene (formed through decarbonylation of stearic acid) was the main product over 5 wt% Cu/Al₂O₃, whereas n-heptadecane (formed through decarboxylation) was the major product on 5 wt% Ni/Al₂O₃ catalyst. Unlike that for the ZrO₂ support,³³ the deoxygenation performance of 5 wt% Cu/Al₂O₃ was higher than that of 5 wt% Ni/Al₂O₃ (yield of total C₁₇ compounds was 78% over the Cu and 36% over the Ni catalyst). 5 wt% Cu/WO₃/ZrO₂ and 5 wt% Ni/WO₃/ZrO₂ exhibited complete conversion of stearic acid. However, cracking was the major reaction over these catalysts. Total C₁₇ compound yield was only 7% over 5 wt% Cu/WO₃/ZrO₂ and 11% over 5 wt% Ni/WO₃/ZrO₂. Denk et al.³⁴ found that Cu increases the activity of Ni for decarbonylation of 1-octadecanal by increasing the electron density of Ni in the bimetallic Ni_xCu_1−x/ZrO₂ catalyst. Thus, the nature of the metal and support and reaction conditions (temperature, pressure and gas environment) influence the catalyst performance and DCO selectivity over non-noble metal catalysts. Highly performing and selective HDO catalysts are still a challenge.

We report here for the first time a highly active and HDO selective, bimetallic Cu–WO_x/Al₂O₃ catalyst for the deoxygenation of oleic acid (OA, a representative fatty acid). The amounts of Cu and W in the catalyst were varied. WO_x promoted the HDO activity of Cu/Al₂O₃. The HDO/DCO product ratio over the catalyst of this study (Cu–WO_x/Al₂O₃) is higher than that known for reduced/sulfided NiMo catalysts. The high HC yield and low carbon loss over the Cu–WO_x/Al₂O₃ catalyst can lead to an improved fatty acid deoxygenation process.

Experimental

Catalyst preparation

Monometallic Cu/Al₂O₃ catalysts with Cu loading in the range of 5–20 wt% and bimetallic Cu–WO_x/Al₂O₃ catalysts with Cu loading at 10 wt% and W in the range of 2–16 wt% were prepared by a wet-impregnation method. γ-Al₂O₃ supplied by Süd-Chemie India Pvt. Ltd., New Delhi, was used as a catalyst support in the preparations.

xCu/Al₂O₃. In the preparation of 5 to 20 wt% Cu deposited on γ-Al₂O₃, the required quantity of Cu(NO₃)₂·3H₂O (99.5%, Loba Chemie) was dissolved in Milli-Q water (10 ml) and 1 g of γ-Al₂O₃ powder was added to it. The suspension was stirred at 40 °C for 4 h. Water was removed over a rotary evaporator (operated at 65 °C). The solid formed was dried at 110 °C for 12 h and calcined at 450 °C (ramp rate = 2 °C min⁻¹) for 4 h. Then, it was reduced at 350 °C for 2.5 h in a flow of hydrogen (20 ml min⁻¹). The catalyst thus formed was labeled as xCu/Al₂O₃, where x refers to the nominal weight per cent of Cu with respect to Al₂O₃.

10Cu–yWO_x/Al₂O₃. The bimetallic 10Cu–yWO_x/Al₂O₃ catalysts were prepared by a sequential wet-impregnation method. Here the number 10 refers to the nominal weight per cent of Cu and y refers to the weight per cent of W. While allowing the Cu content at 10 wt%, the amount of W in the catalyst (y) was varied from 2 to 16 wt%.

Initially, a known quantity of ammonium metatungstate ((NH₄)₆W₁₂O₃₉·xH₂O; 99.9%, Alfa Aesar) was dissolved in 10 ml of Milli-Q water. To that, 1 g of γ-Al₂O₃ powder was added. The suspension was agitated at 40 °C for 16 h. Water was removed over a rotary evaporator. The solid formed was dried at 110 °C for 12 h and calcined at 450 °C (ramp rate = 2 °C min⁻¹) for 4 h. The material thus formed was designated as yWO_x/Al₂O₃.

In the second step, Cu (10 wt%) was deposited on yWO_x/Al₂O₃. The required amount of Cu(NO₃)₂·3H₂O was dissolved in 10 ml of Milli-Q water. To it, 1 g of yWO_x/Al₂O₃ (prepared as above) was added. It was stirred at 40 °C for 4 h. Water was removed over a rotary evaporator (maintained at 65 °C). The solid formed was collected, dried at 110 ° C for 12 h, and calcined at 450 °C for 4 h. It was reduced at 350 °C for 2.5 h in a flow of hydrogen (20 ml min⁻¹). The catalyst thus obtained was labeled as 10Cu–yWO_x/Al₂O₃.

Catalyst characterization procedure

Metal dispersion (D_Cu), active metal surface area (S_Cu; per gram metal) and average crystallite size (CS_Cu) of Cu were determined using a CO-chemisorption technique performed on a Quantachrome Autosorb iQ instrument. In a typical experiment, about 0.2 g of the catalyst was charged into a U-shaped quartz tube. It was degassed at 200 °C (heating rate = 20 °C min⁻¹) in a helium flow (20 ml min⁻¹) for 2 h and then reduced with hydrogen (20 ml min⁻¹) at 350 °C (heating rate = 20 °C min⁻¹) for 2.5 h. Subsequently, it was evacuated (at 350 °C) for 2 h and the sample temperature was lowered to 40 °C (in 1 min) under vacuum. Then, CO gas was introduced and the chemisorption measurements were performed under the following conditions: equilibrium tolerance = 0, equilibrium time = 3 min, adsorption points = 80, 160, 240, 320, 400, 480 and 560 mm Hg, analysis temperature = 40 °C, thermal equilibrium time = 10 min and leak test = 1 min. The following formulae were used to determine the textural properties of supported copper metal.³⁵


Here, N_m is the monolayer CO uptake in units of mol g⁻¹, S is the stoichiometric ratio i.e., number of surface metal atoms covered by each chemisorbed CO molecule, AW is the atomic weight of copper (63.546 g mol⁻¹), L is percent loading of copper, N_a is the Avogadro number (6.023 × 10²³ atoms per mol), A is the cross-sectional area occupied by each active surface Cu atom (6.803 × 10⁻²⁰ m² per atom), f is a particle shape correction factor (considering a spherical/cubic structure, its value is taken as 6) and Z is the density of copper (8.92 × 10⁶ g m⁻³). S_Cu is expressed in units of m² g_Cu⁻¹ and CS_Cu is expressed in nm. The value of the stoichiometric ratio (S) is considered as two (2Cu + CO → Cu(CO)Cu).³⁶

X-ray diffraction (XRD) patterns of reduced catalyst powders were recorded in the 2θ range of 5–90° and with a scan rate of 2° min⁻¹ on an X'Pert Pro Philips diffractometer equipped with a Cu K_α radiation source (λ = 0.15406 nm) and a proportional counter detector. The textural properties of the catalysts were determined at −196 °C using a Micromeritics ASAP 2020 instrument. Nitrogen adsorption–desorption isotherms were collected in the relative pressure (P/P_o) range of 0.01 to 0.99. Prior to nitrogen adsorption, the catalyst samples were outgassed in vacuum at 200 °C for 6 h to achieve clean surfaces. The specific surface area (S_BET) was determined by the Brunauer–Emmett–Teller (BET) method (P/P₀ = 0.01–0.2) and the total pore volume (PV) and pore diameter (PD) were determined using the Barret–Joyner–Halenda (BJH) model.

The acidity of the reduced catalyst was determined by temperature-programmed desorption of ammonia (NH₃-TPD) performed on a Micromeritics Auto Chem 2910 instrument. In a typical experiment, 0.1 g of reduced catalyst was taken in a U-shaped quartz tube. Prior to analysis, pre-treatment was performed in a helium flow (30 ml min⁻¹) at 200 °C for 1 h. Then, the temperature of the sample was brought down to 100 °C and ammonia in helium (10 vol%) was fed for 1 h (at a flow rate of 30 ml min⁻¹). Later, it was flushed with helium (30 ml min⁻¹) for 1 h at the same temperature. Baseline stability was checked prior to the analysis. TPD measurements were recorded in the temperature range of 75 to 550 °C. The plot obtained was deconvoluted (using Origin 6.1 or Pro 8.5) into different peaks corresponding to NH₃ desorption from acid sites of different strengths. The area of the peak and prior calibration enabled quantification of the acid sites.

Transmission electron microscopy (TEM) images of reduced catalyst samples were recorded on a Technai-G2 T20 super twin instrument fitted with a 200 kV field emission gun. The catalyst powder was dispersed in isopropanol, sonicated, drop-casted on a TEM copper grid and dried at 25 °C. Digital micrograph (Gatan) software was used to calculate the size of particles. Around 100–150 particles were considered in particle size estimation.

Hydrogen-temperature-programmed reduction (H₂-TPR) measurements were performed on a Micromeritics Auto Chem 2910 instrument. In a typical procedure, 0.1 g of unreduced catalyst sample was charged in a U-shaped quartz tube. It was heated at 200 °C for 1 h and cooled to 50 °C. The TPR pattern was recorded in the temperature range of 50 to 800 °C while flowing 5 vol% H₂ in Ar (30 ml min⁻¹).

X-ray photoelectron spectroscopy (XPS) studies of the reduced catalysts were conducted on a Thermo Fisher Scientific Instrument, UK (model: K-Alpha⁺) equipped with an Al K_α anode (1486.6 eV) in a transmission lens mode and a multi-channel plate (MCP) detector. The lines of XPS were referenced to C_1s appearing at 284.8 eV. The XP spectra were deconvoluted using XPSPEAK41 software.

Reaction procedure and product analysis

Catalytic tests were performed using a Parr stainless steel batch reactor (300 ml) equipped with a temperature controller. Prior to the reaction, the catalyst was reduced in a flow of hydrogen (20 ml min⁻¹) at 350 °C for 2.5 h. In a typical reaction, known quantities of OA (Sigma Aldrich), n-heptane (solvent, 99.9%, Merck) and the reduced catalyst were placed in an autoclave which was then flushed and pressurized with hydrogen gas (5–20 bar). The temperature of the reactor was raised to 240 to 300 °C and the reaction was conducted for 1 to 5 h. After the reaction, the autoclave was cooled to room temperature (25 °C) and depressurized. The catalyst was separated by centrifugation/filtration. The solvent in the liquid portion was removed over a rotary evaporator and the product isolated was subjected to analysis by titration with NaOH solution (for acid value estimation and OA conversion) and to gas chromatography (GC, for product selectivity) as reported by us earlier.^23,24

Results and discussion

Representative TEM images of 10Cu/Al₂O₃ and 10Cu–4WO_x/Al₂O₃ are shown in Fig. 1. Particle size distribution curves (ESI,† Fig. S1) pointed out a narrow size distribution of Cu particles in the range of 1–5 nm. The average particle size of Cu in 10Cu/A₂O₃ was 2.5 nm and that in 10Cu–4WO_x/Al₂O₃ was 3 nm. Thus, tungsten addition (4 wt%) affected only marginally the Cu particle size. XRD studies revealed that Cu and WO₃ crystallites in the reduced catalysts are below the X-ray detection limit of 3 nm. No separate peaks other than the XRD peaks of γ-Al₂O₃ were detected for 10Cu/Al₂O₃ and 10Cu–4WO_x/Al₂O₃ (Fig. 2).

Fig. 1 TEM images of reduced 10Cu/Al₂O₃ and 10Cu–4WO_x/Al₂O₃ catalysts.

Fig. 2 XRD patterns of γ-Al₂O₃ and reduced 4WO_x/Al₂O₃, 10Cu/Al₂O₃ and 10Cu–4WO_x/Al₂O₃.

CO-chemisorption data of xCu/Al₂O₃ and 10Cu–yWO_x/Al₂O₃ catalysts are presented in Table 1. The percentage of metal dispersion (D_Cu), active metal surface area (S_Cu) and average crystallite size (CS_Cu) of Cu were determined from the monolayer CO uptake values. WO_x/Al₂O₃ did not adsorb CO on its surface. Hence, CO-chemisorption was chosen to measure the dispersion and crystallite size of Cu in copper-loaded samples. The monolayer CO uptake value of different catalysts decreased with increasing Cu and W contents (Table 1). The decrease for a monometallic catalyst (xCu/Al₂O₃) was from 90 to 54 μmol g⁻¹ and for a bimetallic catalyst (10Cu–yWO_x/Al₂O₃) was from 87 to 59 μmol g⁻¹. This variation is mainly attributed to an increase in the crystallite size of Cu (from 5 to 31 nm for xCu/Al₂O₃ and from 10 to 14 nm for 10Cu–yWO_x/Al₂O₃) with increasing metal content. The active metal surface area (per gram of Cu) decreased with increasing metal content from 148 to 22 m² g⁻¹ for xCu/Al₂O₃ and from 71 to 48 m² g⁻¹ for 10Cu–yWO_x/Al₂O₃. Metal dispersion decreased from 23% to 3% for xCu/Al₂O₃ and from 11% to 7% for 10Cu–yWO_x/Al₂O₃. It may be noted that tungsten addition by 4 wt% did not alter the CO-chemisorption data of the Cu catalyst (10Cu/Al₂O₃ and 10Cu–4WO_x/Al₂O₃ had nearly the same metal textural property values; Table 1). But with higher amounts of W (8–16 wt%), the crystallite size of Cu in 10Cu–yWO_x/Al₂O₃ had increased and metal dispersion had decreased (Table 1). A similar observation in variation of CO uptake values and metal crystallite sizes was also reported by Regalbuto et al.³⁷ for Pt/WO_x/SiO₂ catalysts with different amounts of Pt and W. The average crystallite size of Pt increased with increasing Pt and W contents. Zhou et al.³⁸ reported that the dispersion of Pt (measured by CO chemisorption) drops on adding WO_x. The low dispersion in the Pt/WO_x system was attributed to a partial coverage of the Pt surface by the WO_x species.³⁹ They confirmed this by diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy of adsorbed CO. The band at 1837 cm⁻¹ due to bridging CO on the Pt surface (present in Pt/ZrO₂) had disappeared in the case of Pt/WO_x/ZrO₂, inferring a decrease in neighboring Pt atoms on the surface due to partial coverage by the WO_x species.³⁹ Thus, WO_x addition reduced the exposed surface metal sites as a consequence of increased metal crystallite size and/or by surface coverage of metal with WO_x species.

Table 1 N₂-physisorption, CO-chemisorption and NH₃-TPD data of supported Cu catalysts

Catalyst	S BET (m^2 g^−1)	CO-chemisorption				NH3-TPD
		Monolayer CO uptake (μmol g^−1)	S Cu (m^2 gCu^−1)	CSCu (nm)	D Cu (%)	Total acidity (mmol g^−1)	Acid site distribution (mmol g^−1)
							Weak (75–200 °C)	Medium (200–350 °C)	Strong (>350 °C)
γ-Al2O3	352	—	—	—	—	0.99	0.256	0.551	0.183
5Cu/Al2O3	244	90	148	5	23	1.00	0.344	0.556	0.100
10Cu/Al2O3	183	89	73	9	11	1.12	0.314	0.691	0.115
15Cu/Al2O3	171	68	37	18	6	0.93	0.279	0.608	0.043
20Cu/Al2O3	134	54	22	31	3	0.88	0.269	0.571	0.040
10Cu–4WOx/Al2O3	179	87	71	10	11	1.38	0.417	0.638	0.325
10Cu–8WOx/Al2O3	168	79	65	10	10	1.56	0.422	0.919	0.219
10Cu–12WOx/Al2O3	165	63	52	13	8	1.65	0.438	0.996	0.216
10Cu–16WOx/Al2O3	159	59	48	14	7	1.67	0.462	1.077	0.130

---

γ-Al₂O₃ used in this study had a specific surface area (S_BET) of 352 m² g⁻¹. With increasing Cu (5 to 20 wt%) and W (4 to 16 wt%) loading, a marked decrease in S_BET of Al₂O₃ (to 134 m² g⁻¹) was observed (Table 1). The total pore volume (PV) had also decreased (ESI,† Fig. S2 and Table S1), inferring that the pore mouth of γ-Al₂O₃ is blocked partially with Cu and WO_x particles, affecting the textural property values of alumina. The surface density of tungsten (ρ_W; atoms per nm²) was estimated using the following equation:

where y is the percentage loading of tungsten, N_a is Avogadro's number, A_W is the atomic weight of tungsten and S_BET is the specific surface area of the catalyst (in m² g⁻¹), which was found for our catalysts to be in the range of 0.7–3.3 W atoms per nm², which was lower than the theoretical γ-Al₂O₃ monolayer coverage value of 7 W atoms per nm².^40,41

The acidity of the catalysts was determined from NH₃-TPD measurements. All the catalysts showed a broad NH₃ desorption curve in the temperature range of 75 to 550 °C, which was deconvoluted into desorption peaks arising from acid sites of different strengths. Representative NH₃-TPD profiles (continuous lines) and deconvoluted plots (dotted lines) of 10Cu/Al₂O₃ and 10Cu–4WO_x/Al₂O₃ are shown in Fig. 3. Desorption in the temperature range of 75–200 °C is attributed to that arising from the weak acidic sites. Desorption in the temperature range of 200 to 350 °C is attributed to that arising from the acid sites of medium strength, and that above 350 °C is attributed to that of strong acid sites. The overall acidity of the catalyst increased (from 0.99 to 1.12 mmol g⁻¹) with increasing Cu content up to a value of 10 wt% and above that it decreased to 0.88 mmol g⁻¹. A composition with 10 wt% Cu led to a highly acidic catalyst. WO_x addition affected the acidity of the Cu catalyst. It increased the acidity of the catalyst from 1.12 to 1.67 mmol g⁻¹. The weak, medium and strong acid site distribution of 10Cu/Al₂O₃ was 0.31, 0.69 and 0.12 mmol g⁻¹ and that of 10Cu–4WO_x/Al₂O₃ was 0.42, 0.64 and 0.33 mmol g⁻¹, respectively. In other words, a significant increase in weak and strong acid sites and a decrease in medium acid sites was noted on tungsten addition. With a higher amount of tungsten (>4 wt%), the strong acid sites decreased, while the weak and medium acid sites increased (Table 1). A catalyst composition of 10 wt% Cu and 4 wt% W had the highest amount of strong acid sites. At lower concentrations, W is known to be in a highly dispersed state and at higher concentrations, oligomeric tungsten and WO₃ type species are known to be formed.³⁸ While the dispersed tungsten species strongly interacts with the support (Al₂O₃) leading to strong acid sites, the oligomeric and crystalline WO₃ species lead to weak and medium strength acid sites. Si et al.⁴² found by Fourier transform infrared (FT-IR) spectroscopy of adsorbed NH₃ that WO_x–ZrO₂ samples contain Lewis and Brønsted acid sites. NH₃ coordinated to the Lewis acid sites showed characteristic IR bands of NH₃ at 1600 and 1200 cm⁻¹ due to σ_as and σ_s modes. NH₃ coordinated to the Brønsted acid sites (as NH₄⁺) showed bands at 1680 and 1470 cm⁻¹ due to σ_s and σ_as modes. When CuO was deposited (CuO/WO_x–ZrO₂), the intensity of the IR bands due to NH₄⁺ ions decreased, indicating reduction in the concentration of Brønsted acid sites, and those of NH₃ coordinated to Lewis acid sites increased. The higher concentration of weak and medium acid sites and lower concentration of strong acid sites observed in the present study for the copper catalysts with >4 wt% W agrees well with the report of Si et al.⁴²

Fig. 3 NH₃-TPD profiles (continuous black curve) and deconvoluted plots (dotted blue curves) of reduced 10Cu/Al₂O₃ and 10Cu–4WO_x/Al₂O₃ catalysts.

10CuO/Al₂O₃ and 10CuO–4WO_x/Al₂O₃ showed two overlapping reduction peaks in H₂-TPR in the temperature range of 150 to 250 °C (Fig. 4). These are corresponded to reduction of CuO to metallic copper.^43,44 The low temperature peak is attributed to dispersed CuO particles strongly interacting with the support and the high temperature reduction peak is attributed to bulk CuO particles weakly interacting with the support.⁴⁴ Both these reduction peaks shifted to lower temperature when WO_x was present in the catalyst composition (10CuO–4WO_x/Al₂O₃). The low temperature peak shifted from 187 to 178 °C and the high temperature peak shifted from 230 to 221 °C (Fig. 4). Thus, WO_x facilitated the reduction of CuO. The strong interaction between CuO and WO_x particles supported on Al₂O₃ is the reason for the appearance of H₂-TPR peaks at lower temperatures.

Fig. 4 H₂-TPR profiles of as-synthesized 4WO_x/Al₂O₃, 10Cu/Al₂O₃ and 10Cu–4WO_x/Al₂O₃.

No prominent reduction peak in H₂-TPR was observed for 4WO_x/Al₂O₃ in the temperature range of 100 to 800 °C. Thus, WO_x on Al₂O₃ did not reduce in hydrogen consumption below 800 °C. The hydrogen consumption by the catalysts in this temperature range is therefore attributed solely to CuO. The overall hydrogen consumption for CuO reduction was lower for 10CuO–10WO_x/Al₂O₃ (0.273 mmol per g catalyst) than for 10CuO/Al₂O₃ (0.596 mmol per g catalyst). Since the amount of Cu is the same in both the catalysts and all Cu should be reduced below 800 °C, we expect the same hydrogen consumption value for 10CuO/Al₂O₃ and 10CuO–10WO_x/Al₂O₃ (considering that tungsten is not reduced in this temperature range). However, the lower value observed for the latter catalyst tempts us to postulate that a some Cu is perhaps covered with WO_x particles or Al₂O₃ as in the case of Pt/WO_x/ZrO₂ (ref. 39) and hence, not reduced. The active copper content participating in the reactions was lower in the former than in the latter. Based on hydrogen consumption we estimated that only 30 mol% of total Cu in the catalyst was reduced to metallic Cu in 10CuO/Al₂O₃, while it was only 17 mol% in 10CuO–4WO_x/Al₂O₃. The TPR profile was deconvoluted and the low and high temperature peaks were in the proportion of 7% and 93%, respectively, in 10CuO/Al₂O₃ and 62.7% and 37.3%, respectively, in 10CuO–4WO_x/Al₂O₃. Thus, upon modification with WO_x, a greater amount of strongly interacting CuO species had formed in 10CuO–4WO_x/Al₂O₃, while the proportion of weakly interacting CuO particles had decreased. In a related Pt system (4Pt–8WO_x/Al₂O₃),²³ we found earlier that with WO_x addition, the reduction peak of PtO shifts to lower temperatures due to the increased crystallite size of the platinum. WO_x did not show any reduction peak below 800 °C. The present results are in line with these observations. Yang et al.⁴⁵ found that when the tungsten loading was higher (10 wt%, for example, as in the case of xNi–10W/Al-MCM-41), reduction of both NiO (368–483 °C) and WO_x (500–735 °C) was observed. Unlike that observed for our system, the reduction peaks of NiO shifted to a higher temperature, revealing the formation of weakly interacting nickel species with the support. Perhaps, the difference in W content is the reason for the different reduction behaviors and metal–support interactions. A lesser amount of W leads to dispersed WO_x species which require a higher temperature (>800 °C) as seen in the present study. Further, Yang et al.⁴⁵ reported formation of a very active Ni₁₇W₃ species in their catalyst composition. However, similar such Cu–W and Cu–Al phases were not detected by XRD for our catalyst compositions (Fig. 2).

The oxidation states of copper and tungsten in reduced 10Cu/Al₂O₃ and 10Cu–4WO_x/Al₂O₃ catalysts were determined using XPS technique. The spectra of Cu in the 2p core level region and of W in the 4f core level region are depicted in Fig. 5. Spectral fitting of Cu 2p peaks revealed the presence of copper in +2, +1 and zero oxidation states. The shake-up satellite peak centered at around 944 eV confirmed the presence of Cu²⁺ species.^46,47 Thus, not all copper was present in the metallic state in the measured catalysts. The intensity of the shake-up satellite was low for 10Cu–4WO_x/Al₂O₃. The relative percentages of surface Cu species were calculated from the ratio of the areas of corresponding characteristic peaks and listed along with the binding energy values in Table 2. For 10Cu/Al₂O₃, the 2p_3/2 and 2p_1/2 lines of Cu²⁺ species occurred at 936.4 and 956.2 eV. Those of the Cu¹⁺ species appeared at 934.8 and 954.6 eV and those of the Cu⁰ species appeared at 932.7 and 952.8 eV.⁴⁸ These XP lines of Cu (2+, 1+ and 0) for 10Cu–4WO_x/Al₂O₃ appeared at lower binding energy values (by 0.2–0.6 eV, Table 2) indicating that copper in the bimetallic catalyst is electron rich and hence more metallic in nature than that in the monometallic catalyst. The relative concentration ratio of 0, +1 and +2 Cu species is 33 : 37 : 30 (for 10Cu/Al₂O₃) and 56 : 26 : 18 (for 10Cu–4WO_x/Al₂O₃). Thus, the WO_x-modified catalyst contained a higher proportion of reduced copper than the unmodified catalyst. The reduction curves in H₂-TPR (appearing at lower temperature) have also indicated that copper on WO_x-promoted Al₂O₃ reduces more easily than on pure Al₂O₃.

Fig. 5 XPS profiles of reduced 10Cu/Al₂O₃ in the Cu 2p core level region and 10Cu–4WO_x/Al₂O₃ in the Cu 2p and W 4f core level regions. The deconvoluted peaks of Cu⁰ (green), Cu¹⁺ (blue) and Cu²⁺ (pink) as well as of W⁵⁺ (blue) and W⁶⁺ (red) are indexed.

Table 2 XPS data of supported Cu catalysts

Catalyst	Cu 2p peak positions (eV)						Relative intensity (%)			W 4f peak positions (eV)				Relative intensity (%)
	Cu^0		Cu^1+		Cu^2+		Cu^0	Cu^1+	Cu^2+	W^5+		W^6+		W^5+	W^6+
	2p3/2	2p1/2	2p3/2	2p1/2	2p3/2	2p1/2				4f7/2	4f5/2	4f7/2	4f5/2
10Cu/Al2O3	932.7	952.8	934.8	954.6	936.4	956.2	33	37	30	—	—	—	—	—	—
10Cu–4WOx/Al2O3	932.2	952.2	934.2	954.0	936.2	956.0	56	26	18	34.8	37.0	35.9	38.0	60	40

---

This difference in the concentration and electronic structure of Cu speciation in supported Cu catalysts can bring about variation in catalytic activity.^22–24 The H₂-TPR curves of 10Cu/Al₂O₃ and 10Cu–4WO_x/Al₂O₃ (Fig. 4) showed complete reduction of copper at 230.1 and 221.4 °C, respectively. However, detection of Cu¹⁺ and Cu²⁺ species in XPS for the samples reduced at 350 °C for 2.5 h under a hydrogen flow of 20 ml min⁻¹ reveals that the catalyst samples were exposed to air during handling, allowing some of the copper to oxidize to +1 and +2 states during sample preparation for the XPS measurement. Alternatively, copper covered with WO_x and Al₂O₃ particles is perhaps not fully reduced and showed peaks corresponding to +1 and +2 species. Further, Cu–WO_x and Cu–Al₂O₃ phases in the catalysts can explain the presence of Cu¹⁺ and Cu²⁺ species in the reduced catalysts. Such phases were, however, not detected in the XRD (Fig. 2). Thus, even if they were there, they might be present as amorphous phases not showing XRD peaks.

Curve fitting of the W 4f XP spectrum confirmed the presence of tungsten in both +5 and +6 oxidation states. The 4f_7/2 and 4f_5/2 lines of W⁵⁺ species appeared at 34.8 and 37.0 eV and those of W⁶⁺ species appeared at 35.9 and 38.0 eV.^49–51 Neat WO_x/Al₂O₃ contained W in the +6 oxidation state only. However, in the presence of Cu and in a hydrogen environment, some of the tungsten oxide reduced to acidic W⁵⁺O_xH species. These reduced tungsten oxide species are known to activate fatty acids (OA, for example) and alter the deoxygenation mechanism toward HDO.^21–24 The relative surface concentration of W⁵⁺ : W⁶⁺ was in the ratio of 60 : 40. The reduction of tungsten from the +6 to +5 oxidation state was found higher in Pt (ref. 23) than in Cu–WO_x/Al₂O₃ catalyst. Higher temperature (>800 °C) is needed to reduce “neat” WO₃. However, tungsten oxide in the presence of a metal and in a hydrogen environment can reduce at lower temperatures. XPS confirmed such a reduction of W⁶⁺ to W⁵⁺ species in 10Cu–4WO_x/Al₂O₃ treated with hydrogen at 350 °C for 2.5 h. However, H₂-TPR did not show the corresponding tungsten reduction peak (Fig. 4) perhaps due to the low sensitivity of the thermal conductivity detector used or the baseline of the curve covered the broad and tiny reduction peaks of tungsten.

Catalytic activity

The catalytic activity data of Al₂O₃-supported xCu and 10Cu–yWO_x catalysts for deoxygenation of oleic acid (OA) are presented in Table 3. Deoxygenation of OA yielded n-octadecane (C₁₈) formed by the HDO path and n-heptadecane (C₁₇) formed by the DCO path. Octadecanal and octadecanol were the intermediate (incomplete deoxygenation) products (referred to as others). C_10–16 HCs were the cracking products. A control experiment at 300 °C and 20 bar H₂ pressure after 5 h of reaction revealed that very little deoxygenation occurred (OA conversion = 4 mol%) in the absence of a catalyst (Table 3, run no. 1). Over 8WO_x/Al₂O₃, the conversion of OA at these conditions was only 8 mol% (run no. 2). However, with the xCu/Al₂O₃ catalyst, OA conversion increased to 38–97 mol% depending on the copper content (x) (run no. 3–6). 10Cu/Al₂O₃ enabled the highest OA conversion of 97 mol% after 5 h of reaction (run no. 4). Others (octadecanal and octadecanol) were the major products over this catalyst. It gave a higher amount of C₁₈ than the other Cu compositions. The turnover frequency (TOF = moles of C₁₈ formed per mole of surface Cu atoms (estimated from metal dispersion obtained from CO-chemisorption) per hour) was the highest (19 h⁻¹) for the 10Cu/Al₂O₃ catalyst.

Table 3 Catalytic hydrodeoxygenation activity data of Cu/Al₂O₃ and Cu–WO_x/Al₂O₃^a

Run no.	Catalyst	Reaction temp. (°C)	H2 pressure (bar)	Reaction time (h)	OA conv. (mol%)	TOF^b (h^−1)	Product selectivity (%)
							C18	C17	C10–16	Others
a Reaction conditions: OA = 2 g, n-heptane (solvent) = 30 g, catalyst = 0.2 g. b Turnover frequency (TOF) = moles of C18 formed per mole of surface Cu atoms (estimated from CO-chemisorption) per hour.
1	Nil	300	20	5	4.0	—	5.0	8.0	0.4	86.6
2	8WOx/Al2O3	300	20	5	9.0	—	9.0	43.2	0.8	47.0
3	5Cu/Al2O3	300	20	5	38.0	2	13.3	12.7	0.5	73.5
4	10Cu/Al2O3	300	20	5	97.0	19	48.4	12.5	0.1	39.0
5	15Cu/Al2O3	300	20	5	87.0	14	32.5	3.6	0.1	63.8
6	20Cu/Al2O3	300	20	5	67.0	10	20.0	4.4	0.1	75.5
7	10Cu–2WOx/Al2O3	300	20	1	35.0	7	10.4	8.9	0.2	80.5
8	10Cu–4WOx/Al2O3	240	20	1	10.0	1	2.9	1.4	0.2	95.5
9	10Cu–4WOx/Al2O3	260	20	1	22.0	4	8.0	3.7	0.3	88.0
10	10Cu–4WOx/Al2O3	280	20	1	72.0	114	77.6	3.6	0.7	18.1
11	10Cu–4WOx/Al2O3	300	5	1	39.0	10	13.7	7.6	0.4	78.3
12	10Cu–4WOx/Al2O3	300	10	1	51.0	27	25.6	8.9	0.5	65.0
13	10Cu–4WOx/Al2O3	300	20	1	97.0	182	91.9	3.4	0.6	4.0
14	10Cu–4WOx/Al2O3	300	30	1	100	190	93.0	4.5	0.4	2.1
15	10Cu–8WOx/Al2O3	300	20	1	91.0	191	93.2	4.0	0.3	2.5
16	10Cu–8WOx/Al2O3	300	20	5	100.0	40	89.4	2.0	0.5	8.1
17	10Cu–12WOx/Al2O3	300	20	1	56.0	132	84.0	5.4	0.6	10.0
18	10Cu–16WOx/Al2O3	300	20	1	53.0	126	85.0	4.5	0.5	10.0

---

To monitor the effect of WO_x on the catalytic activity of Cu, bimetallic catalysts (with 10 wt% Cu and varying amounts (y) of W) were evaluated for 1 h at 300 °C and 20 bar H₂ pressure (run no. 7, 13, 15, 17 and 18). OA conversion increased with increasing W content up to 4 wt% and above that the OA conversion decreased. 10Cu–4WO_x/Al₂O₃ exhibited OA conversion of 97% at the end of 1 h, which the monometallic catalyst (10Cu/Al₂O₃) yielded only after 5 h (compare run no. 13 with 4). At similar conversions (ca. ∼97%), C₁₈ is the major product over the bimetallic catalyst (run no. 13), while significant amounts of others (39.0%) and C₁₇ (12.5%) formed on the monometallic Cu catalyst (Fig. 6). Thus, WO_x enhanced the catalytic HDO activity and selectivity of Cu/Al₂O₃.

Fig. 6 Plot showing the influence of WO_x on product selectivity at similar conversions of OA (97.0%) over 10Cu/Al₂O₃ and 10Cu–4WO_x/Al₂O₃ catalysts. Reaction conditions: see Table 3, run no. 4 and 13.

Huang et al.⁵² employed a coprecipitation approach for preparing bifunctional Ru–W/SiAl catalysts for the hydrogenolysis of phenols and phenyl ethers to arenes. Coprecipitation is a simpler method for catalyst preparation. However, we followed sequential wet impregnation, as this method lends most of the Cu to locate on WO_x-modified Al₂O₃ while coprecipitation can lead to Cu with little contact with WO_x. Note that the catalytic activity tests (Table 3) revealed that Cu located on WO_x-modified Al₂O₃ shows higher HDO selectivity (C₁₈) while that on unmodified Al₂O₃ is active for hydrogenation and DCO (C₁₇) reactions.

Having found that 10Cu–4WO_x/Al₂O₃ is the best HDO catalyst, it was evaluated at different temperatures and pressures. OA conversion and C₁₈ selectivity increased with increasing reaction temperature from 240 to 300 °C (run no. 8, 9, 10 and 13). Also, with increasing H₂ pressure, OA conversion and HDO selectivity increased (run no. 11–14). At 300 °C and 30 bar H₂, the bimetallic catalyst (10Cu–4WO_x/Al₂O₃) showed complete conversion of OA in 1 h with 93% selectivity for the C₁₈ product (run no. 14). TOF was 190 h⁻¹. The HDO/DCO product ratio over this catalyst was 93.0/4.5. To the best of our knowledge, this is the highest HDO selectivity reported to date in deoxygenation of vegetable oils. Non-noble metal NiMo and NiW catalysts are known to yield an HDO : DCO product ratio in the range of 69 : 21 to 20 : 80.^14,15 Pt–WO_x/Al₂O₃ (ref. 23) and Pt–MoO_x/ZrO₂ (ref. 24) gave an HDO : DCO product ratio of 88 : 9 and 91 : 7.3, respectively. The highest activity and HDO selectivity over the non-sulfided, non-noble metal catalyst of the present study is therefore expected to lead to an eco-friendly, economically beneficial fatty acid deoxygenation process producing renewable HCs (green diesel). The optimal reaction conditions for the HDO reaction were 300 °C and 30 bar H₂ (Table 3), but when comparing the catalytic performances of the catalysts with varying W loadings (entries 7, 13, 15, 17 and 18) we employed 20 bar H₂ instead of 30 bar because better differentiation can be obtained when the catalysts are compared at conditions below the optimum level.

Irrespective of reaction conditions and composition, a correlation existed between OA conversion and C₁₈ yield over 10Cu–yWO_x/Al₂O₃ catalysts (Fig. 7). The higher the OA conversion, the higher was the selectivity/yield of C₁₈ product. Any deviation from the correlation is perhaps due to changes in the crystallite size and acidity of the catalysts. In fact, we found a correlation (Fig. 8) between the TOF data (Table 3) and the average crystallite size of Cu as well as the acidity of the catalysts (Table 1). There seems to be an optimum value for the crystallite size of Cu (10–10.5 nm) and acidity (1.56 mmol g⁻¹) to get the highest TOF value. The TOF of the catalyst was lower above and below that critical value. The catalyst with 10 wt% Cu and 4–8 wt% W had the optimum composition. Too low or high content of W leads to catalysts with inferior catalytic performance.

Fig. 7 Correlation between OA conversion and yield of C₁₈ over 10Cu–yWO_x/Al₂O₃ catalysts. Data taken from Table 3 (run no. 7–15, 17 and 18).

Fig. 8 Dependence of TOF on the average crystallite size of Cu for xCu/Al₂O₃ and 10Cu–yWO_x/Al₂O₃ catalysts (left) and on the acidity of the 10Cu–yWO_x/Al₂O₃ catalysts (right). Data taken from Tables 1 and 3.

10Cu–4WO_x/Al₂O₃ was stable and reusable. After the catalytic run, it was separated from the liquid product by centrifugation/filtration. Then, it was washed with acetone, dried (110 °C), reduced in hydrogen (350 °C) and used in the recycle run (conducted at 300 °C and 30 bar H₂ for 1 h). Quantitative conversion of OA with C₁₈, C₁₇, C_10–16 and other products selectivity of 93.5%, 4.2%, 0.3% and 2.0%, respectively, was yielded. The spent catalyst showed similar structural characteristics (XRD and CO-chemisorption) to that of the fresh catalyst, confirming its stability in the reaction.

The XPS study pointed out that the proportion of metallic copper (Cu⁰) was higher in the WO_x-promoted bimetallic Cu catalyst (56%) than in the unpromoted monometallic Cu catalyst (33%) (Table 2). In the presence of copper, a number of W species (60%) were reduced from the +6 to the +5 oxidation state, creating additional acidity (Table 1). For the related Pt catalysts,^22–24 the reduced WO_x species (instead of Al₂O₃) activate OA and direct the reaction path from DCO to HDO. If the role of Cu–WO_x is the same as that of Pt–WO_x in our earlier studies,^22–24 the HDO/DCO selectivity would have remained unchanged. The higher HDO/DCO selectivity obtained over the Cu–WO_x catalyst points out a mechanism other than that over the Pt–WO_x catalyst (OA adsorption on reduced WO_x, resulting in the formation of an ester between WO_x–H and OA and following that the hydrogenation of the C=C bond and successive deoxygenation forming octadecane with no other intermediates like aldehyde and alcohol). In the well-established reaction network (as on NiMo catalysts), the reduction of OA to aldehyde and its rapid equilibrium with the corresponding alcohol, the parallel decarboxylation of aldehyde to heptadecane and dehydration of alcohol to octadecane is the reaction pathway. Observation of a significant amount of others (aldehyde + alcohol) over the monometallic xCu/Al₂O₃ catalysts (Table 3, run no. 3–6), infers that the reaction on Cu–WO_x may proceed the same as that on the NiMo catalysts. But then, the difference is that the dispersed, acidic (mainly moderate and strong acidity), reduced WO_x–H facilitates OA adsorption (as in the case of Pt–WO_x) and may increase the dehydration of the intermediate alcohol. The electronically richer Cu⁰ may accelerate the hydrogenation of the intermediate octadecene to octadecane. The dependence of TOF on the Cu crystallite size (Fig. 8) suggests that hydrogen adsorption and hydrogenation could be the rate-limiting step of the deoxygenation reaction. Higher acidity and reduced Cu⁰ species are, therefore, responsible for the higher HDO activity of the Cu–WO_x/Al₂O₃ catalysts. Formation of considerable amounts of intermediate esters by reaction of the intermediate aldehyde and alcohol was reported for Ni catalysts and mainly for the deoxygenation under solvent-free conditions.¹⁵ The absence of these ester intermediate compounds in the present work is due to the different catalytic behavior of copper with respect to nickel or due to the presence of solvent. Detailed kinetic and reaction mechanism studies are needed for further confirmation, which will be the topic of our further study.

Conclusions

Bimetallic WO_x-promoted Cu/Al₂O₃ was found to be more active and HDO selective than the monometallic Cu/Al₂O₃ catalyst in converting OA into renewable fuel-grade HCs. A catalyst with 10 wt% Cu and 4 wt% W exhibited complete conversion of OA with 93% selectivity to n-octadecane (C₁₈) at 300 °C, 30 bar H₂, reaction time of 1 h and reactant oleic acid-to-catalyst weight ratio of 2 g/0.2 g. The HDO product yield over this catalyst is the highest known to date for a supported non-noble metal catalyst. The crystallite size of Cu, reducibility of metal and acidity of the catalyst influenced the catalytic HDO activity and selectivity. A bimetallic composition having copper with an optimum crystallite size of about 10 nm and acidity of 1.56 mmol g⁻¹ was found to be the highest HDO product yielding catalyst.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

SD thanks the Council of Scientific and Industrial Research (CSIR), New Delhi, for the CSIR-Bhatnagar Fellowship (SSB 000926; P90807). The authors thank Dr. Amitava Das and Dr. P. S. Subramanian (CSIR-Central Salt & Marine Chemicals Research Institute, Bhavnagar, India) for allowing us to use the nitrogen-physisorption facility.

Notes and references

1. G. W. Huber, S. Iborra and A. Corma, Chem. Rev., 2006, 106, 4044.
2. G. W. Huber and A. Corma, Angew. Chem., Int. Ed., 2007, 46, 7184.
3. T. V. Choudhary and C. B. Phillips, Appl. Catal., A, 2011, 397, 1.
4. C. Zhao, T. Bruck and J. A. Lercher, Green Chem., 2013, 15, 1720.
5. Biomass to Biofuels: Strategies for Global Industries, ed. A. A. Vertés, N. Qureshi, H. P. Blaschek and H. Yukawa, John Wiley & Sons Ltd., 2010.
6. S. Lestari, P. Mäki-Arvela, H. Bernas, O. Simakova, R. Sjöholm, J. Beltramini, G. Max Lu, J. Myllyoja, I. Simakova and D. Y. Murzin, Energy Fuels, 2009, 23, 3842.
7. B. Donnis, R. G. Egeberg, P. Blom and K. G. Knudsen, Top. Catal., 2009, 52, 229.
8. http://www.nesteoil.com .
9. J. A. Petri and T. L. Marker, US Pat., 7511181B2, 2009 – UOP LLC, US.
10. T. L. Marker, P. Kokayeff, S. F. Abdo and T. N. Kalnes, US Pat., 8003836B2, 2011 – UOP LLC, US.
11. http://www.uop.com/pr/release/PR.EniEcofiningFacility.pdf .
12. http://www.tyson.com/RenewableEnergy/Iniotiative/Conoco/Default.aspx .
13. D. Kubička and H. Horáček, Appl. Catal., A, 2011, 394, 9.
14. R. W. Gosselink, S. A. W. Hollak, S.-W. Chang, J. van Haveren, K. P. de Jong, J. H. Bitter and D. S. van Es, ChemSusChem, 2013, 6, 1576.
15. C. Kordulis, K. Bourikas, M. Gousi, E. Kordouli and A. Lycourghiotis, Appl. Catal., B, 2016, 181, 156.
16. L. Simakova, O. Simakova, P. Mäki-Arvela, A. Simakova, M. Estrada and D. Y. Murzin, Appl. Catal., A, 2009, 355, 100.
17. M. Amadi, A. Nambo, J. B. Jasinski, P. Ratnasamy and M. A. Carreon, Catal. Sci. Technol., 2015, 5, 380.
18. M. V. Tsodikov, A. V. Chistyakov, M. A. Gubanov, P. A. Zharova, S. S. Shapovalov, A. A. Pasynskii, V. V. Kriventsov and I. I. Moiseev, Russ. Chem. Bull., 2015, 64, 2062.
19. L. Boda, G. Onyestyák, H. Solt, F. Lónyi, J. Valyon and A. Therrnesz, Appl. Catal., A, 2010, 374, 158.
20. S. Janampelli and S. Darbha, ChemistrySelect, 2017, 2, 1895.
21. S. Janampelli and S. Darbha, Catal. Today, 2018, 309, 219.
22. S. Janampelli and S. Darbha, Mol. Catal., 2018, 451, 125.
23. S. Janampelli and S. Darbha, Energy Fuels, 2018, 32, 12630.
24. S. Janampelli and S. Darbha, Catal. Commun., 2019, 125, 70.
25. M. Peroni, G. Maneino, E. Barath, O. Y. Gutierrez and J. A. Lercher, Appl. Catal., B, 2016, 180, 301.
26. S. Foraita, Y. Liu, G. L. Haller, E. Baráth, C. Zhao and J. A. Lercher, ChemCatChem, 2017, 9, 195.
27. A. Kiméné, R. Wjcieszak, S. Paul and F. Dumeignil, J. Chem. Technol. Biotechnol., 2019, 94, 658.
28. H. Imai, T. Kimura, K. Terasaka, X. Li, K. Sakashita, S. Asaska and S. S. Al-Khattaf, Catal. Today, 2018, 303, 185.
29. R. G. Kurushkkin, O. A. Bulavchenko, V. V. Kaichev and V. A. Yakovlev, Appl. Catal., B, 2015, 161, 531.
30. E. Kordouli, L. Sygellou, C. Kordulis, K. Bourikas and A. Lycourghiotis, Appl. Catal., B, 2017, 209, 12.
31. E. Kordouli, b. Pawelec, K. Bourikkas, C. Kordulis, J. L. G. Fierro and A. Lycourghiotis, Appl. Catal., B, 2018, 229, 139.
32. J. Wu, J. Shi, J. Fu, J. A. Leidl, Z. Hou and X. Lu, Sci. Rep., 2016, 6, 27820.
33. A. S. Berenblyum, R. S. Shamsiev, T. A. Podoplelova and V. Y. Danyushevsky, Russ. J. Phys. Chem. A, 2012, 86, 1199.
34. C. Denk, S. Foraita, Y. Liu, K. Stoerzinger, Y. Liu, E. Baráth and J. A. Lercher, Catal. Sci. Technol., 2019, 9, 2620.
35. https://nptel.ac.in/courses/103103026/14 .
36. Gas sorption system operating manual, model 6, version 4.0, Quantachrome Instruments, Boynton Beach, FL, USA, 2015.
37. J. R. Regalbuto, T. H. Fleisch and E. E. Wolf, J. Catal., 1987, 107, 114.
38. W. Zhou, J. Luo, Y. Wang, J. Liu, Y. Zhao, S. Wang and X. Ma, Appl. Catal., B, 2019, 242, 410.
39. W. Zhou, Y. J. Zhao, S. P. Wang and X. B. Ma, Catal. Today, 2017, 298, 2.
40. D. G. Barton, S. L. Soled, G. D. Meitzner, G. A. Fuentes and E. Iglesia, J. Catal., 1999, 181, 57.
41. S. Garcia-Fernández, I. Gandarias, J. Requies, M. B. Güemez, S. Bennici, A. Auroux and P. L. Arias, J. Catal., 2015, 323, 65.
42. Z. Si, D. Weng, X. Wu, J. Li and G. Li, J. Catal., 2010, 271, 43.
43. D. L. Hoang, T. T. H. Dang, J. Engeldinger, M. Schneider, J. Radnik, M. Richter and A. Martin, J. Solid State Chem., 2011, 184, 1915.
44. G. Vidya Sagar, P. V. Ramana Rao, C. S. Srikanth and K. V. R. Chary, J. Phys. Chem. B, 2006, 110, 13881.
45. R. Yang, X. Du, X. Zhang, H. Xin, K. Zhou, D. Li and C. Hu, ACS Omega, 2019, 4, 10580.
46. T. Venkov, M. Dimitrov and K. Hadjiivonov, J. Mol. Catal. A: Chem., 2006, 243, 8.
47. J. Morales, A. Caballero, J. P. Holgado, J. P. Espinós and A. R. González-Elipe, J. Phys. Chem. B, 2002, 106, 10185.
48. D. Mukherjee, R. Singuru, P. Venkataswamy, D. Damma and B. M. Reddy, ACS Omega, 2019, 4, 4770.
49. T. Y. Kim, D. S. Park, Y. Choi, J. Baek, J. R. Park and J. Yi, J. Mater. Chem., 2012, 22, 10021.
50. S. Zhu, X. Gao, Y. Zhu and Y. Li, J. Mol. Catal. A: Chem., 2015, 398, 391.
51. C. Liu, C. Zhang, S. Hao, S. Sun, K. Liu, J. Xu, Y. Zhu and Y. Li, Catal. Today, 2016, 261, 116.
52. Y.-B. Huang, L. Yan, M.-Y. Chen, Q.-X. Guo and Y. Fu, Green Chem., 2015, 17, 3010.

---

Footnote

† Electronic supplementary information (ESI) available: Supporting figures and table. See DOI: 10.1039/c9cy01939a

---